Title: Machine Learning-Based Mix Design Tools to Minimize Carbon Footprint and Cost of UHPC

Author(s): Tavares

Publication: Web Session

Volume: ws_S22_Tavares.pdf

Issue:

Appears on pages(s):

Keywords:

DOI:

Date: 3/28/2022

Abstract:
The last couple of decades have been characterized by a global awareness on rising greenhouse gas emissions, with the Intergovernmental Panel on Climate Change calling for 50-85% reductions in these emissions by 2050 to prevent threatening climate changes. The building and construction industries have been among the leading consumers of material by mass for almost 100 years, with approximately 5% of all anthropogenic global CO2 emissions being originated from Portland cement manufacture. With global population growth and rapid development of third world countries, it is very unlikely that advancements in the process efficiency of material manufacture will occur fast enough to meet emission reduction needs. Thus, new metrics that account for material performance and environmental impact concurrently during design and material selection stages is of great importance and represents a great opportunity for game-changing methods to reduce emissions. This study presents an innovative approach to design and characterize concrete materials with strengths up to UHPC levels, using a framework that allows one to efficiently design concrete mixtures considering multi-objective performance levels. A strategic framework of experimental data collection is coupled with machine learning models to generate performance and emission density diagrams (PDDs and EDDs, respectively). These diagrams intend to provide a flexible tool to evaluate performance, durability, and environmental impacts concurrently, without adding exhausting experimental campaigns. Addressing this issue is critical for public education and industry engagement on next-gen concrete materials such as UHPC, considering the very high cement content often required to achieve the outstanding properties of this material.